DartFS: Storage and Computing Information

DartFS:

• DartFS is the primary storage solution at Dartmouth.
• A robust system backed by a 3.9-petabyte Dell/EMC PowerScale storage array. The supercomputing infrastructure seamlessly integrates with DartFS through Kerberized NFSv4, ensuring secure access and data protection.
• Users can also leverage the (SMB) protocol to conveniently mount the storage, offering flexibility in accessing and managing stored data.

Discovery:

• The Discovery cluster consists of over 5,000 cores and serves as a vital resource for the Dartmouth research community.
• Operating on the CentOS 7 Linux distribution, this high-performance computing infrastructure features compute nodes equipped with 8GB of memory per core, in addition to 1TB solid-state drive (SSD) local scratch space.
• The nodes are available in configurations of 16, 40, and 64 cores catering to diverse computation requirements.
• The system is interconnected through a 10 Gigabit Ethernet network, with the majority of compute nodes benefiting from a high-speed, 100 Gigabit InfiniBand backplane.

Polaris:

• Polaris is a shared memory computer (SMP) with 40 cores, 64-bit Intel processors, and 1 Terabyte of memory.
• Operating on Red Hat Enterprise Linux 7.
• It is used to run computationally intensive programs such as Mathematica and statistical applications, or programs that require a lot of shared memory.
• It has 5 TB of local scratch space available. Access is by SSH (remote login).

Andes:

• Andes is a shared memory computer (SMP) with 60 physical cores (120 logical cores with HyperThreading turned on), 64-bit Intel processors, and 512 GB of memory. 
• Operating on Red Hat Enterprise Linux 7.
• It is mostly used to run statistical packages and scientific applications that need large amounts of memory and scratch space.  5 TB of fast scratch space is available.
• Access is by SSH (remote login).

Access Controls:

• Dartmouth's data center is housed within a secure facility, requiring dual-key card authentication and individualized personal identification numbers (PINs) for entry clearance.
• Users of DartFS are allocated individual home directories and group laboratory spaces.
• Access management is facilitated through Active Directory (AD) integration, and a comprehensive Access Control List (ACL) structure ensures granular control over permissions, providing a secure environment for data storage and management.

Data Encryption:

• Users will only be able to access the cluster via the on-campus network or by accessing Dartmouth VPN. To login to the cluster, users must use a client that supports SSH or SFTP.
• Protected by nfs4 facls. 
• Encryption at rest data.

Support Team:

The Research Computing team offers the following services to its users:

• Comprehensive user assistance and educational resources for effective cluster and HPC systems utilization.
• Expertise in systems administration, including software deployment and maintenance.
• Guidance in code development, troubleshooting, parallelization, and performance optimization.
• Provision of essential infrastructure, such as electrical power, cooling, and data center facilities.
• Robust network capabilities.
• Administration of system software, clustering tools, configuration, and access to standard compilers and research applications (e.g., MATLAB, Fortran 90/95, intel compilers).
• Expedited repair services to minimize downtime.
• Access to local, NFS, and temporary storage solutions for data management for computational executions.